Team:Peking/Design

Design

Our project design is based on DNA storage and base editing technology, combined with appropriate scoring and visualization strategies to achieve the combination of technology and art(Figure1).

Figure1

In the first step, we encode music into DNA through appropriate encoding methods. We use two sets of coding methods. The first coding method is to use continuous coding based on single tone. We use 12 bases to edit the time, pitch, volume and position of each tone. This coding method can achieve effective storage. However, there are several problems as follows: first, this coding method leads to different sounds, because of its pitch, volume and so on In this case, if we want to cause mutations, we need a large number of gRNA sequences to cause limited mutations; second, such coding mode, once the excision type mutation occurs, will lead to a large number of changes in the whole music, which is likely to become noise. So we choose the second coding rules, which are based on the rules of instruction set. First, we need to have an initial piece of music, which may be an existing music, such as a classic song or a sequence of a segment. Then, we operate the music through the base coding instruction. According to the MIDI coding rules, we get the music every 128 minutes. For example, when giving instructions, it will locate a sixteenth note, and then operate on the note represented by the note, such as tonality conversion, pitch rise and fall, rhythm re-division, etc. Base pairs contain locators, type symbols, and specific operations, such as tonality conversion, pitch rise, pitch drop, rhythm conversion, etc. This coding method limits the type of operation, so it will not produce dramatic changes. Instead, it would generate discordant sounds. In addition, since every 128 minutes of music sampling, each tone will be converted to produce the final result after multiple rounds, which not only improves the diversity of mutation, but also ensures the stability of the mutation. In addition, in the gene design, we use an anchor link coding sequence to achieve multiple mutations of the same gene(Figure2).

Figure2

In the second step, we mutate the coding gene. There are two ways of mutation, one is computer simulation mutation, the other is to introduce into E. coli to achieve mutation. For the latter, we use two mutation systems: one is base editor system based on dCas9 to realize directed mutation; the other is EvolvR system based on nCas9, which can realize random mutation. In addition, the selection of base editor corresponds to our coding solution. In response to the first generation of coding, we choose base editors such as ABE and CBE. These base editors can edit and delete some bases, but the editing window is narrow. Besides, they tend to introduce indels, leading to dramatic changes in music and even producing white noise. So we replaced the nCas9 used in these base editors with dCas9. This kind of base editor attenuates music mutation. To solve the problem of requiring too much sgRNA sequences, we found this special adenine deaminase named DddA, which edits dsDNA without having to unwind it. In this way we could separate the sequence to be edited from guiding sequence, further decreasing the number of sgRNA needed for massive parallel single-site mutation.

After the mutation, we decode the mutated gene and restore it to music. Then, in the third step, we use our own scoring system to score the music after the mutation. The scoring process is divided into six steps: the first step is to change the tonality of music to C major; the second part is to divide the bars and phrases based on step1; the third step is to score the single bar; the fourth step is to score the connection between the small sections; the fifth step is to score the connection between the phrases, and the final score is obtained by integrating the above scores. After that, we choose the optimal mutation strategy according to the score, and cycle the steps above until we get music with excellent scores.

In addition, we use DNA as a bridge to realize the visualization of music. We transform the DNA corresponding to music into beautiful patterns through different decoding methods to realize the transformation from sound to painting.